Preparing Fire Emissions for CMAQ

October 7, 2014

Objective

Provide options for preparing the PMDETAIL fire inventories for CMAQ 2011 modeling.

Description

Task 1 of the 3SAQS 2015 Work Plan calls for the development of a year 2011 CMAQ modeling platform. Fires (wild, agricultural, and prescribed) are significant air pollution sources that require special treatment in the air quality modeling system due to their unique physical and chemical characteristics. The PMDETAIL fire inventory includes information on specific fire events, including the size of the fires (daily area burned and pollution fluxes), timing, spatial (horizontal and vertical extent), and chemical parameters of the fires. These parameters are read in by software that translates the fire inventories into a format and terms that can be used by an air quality model for simulating the air quality impacts of the fires. ENVIRON both developed software for preprocessing the PMDETAIL inventories for input to CAMx and modified CAMx for modeling these data. Similar software and modifications are needed before the PMDETAIL fires can be simulated in CMAQ.

The two parameters of the PMDETAIL fire inventories that require special treatment include the horizontal spatial allocation (gridding) and vertical plume distribution (plume rise).

Gridding the fire inventories

The PMDETAIL fire inventory provides coordinates for daily fire event centroids. When allocating the fire emissions to modeling grid cells, care must be taken to consider the ratio between the fire size (acres) and the grid cell area. The reason for this consideration is that as grid resolutions increase, the fire emissions become concentrated in a single grid cell, leading to overestimates of the local air quality impacts from the fires. The extremely concentrated emissions can also cause problems for the numerical solvers in the air quality models. The table below shows the relationship between grid resolution, grid cell area, and acreage. Many of the large wildfires in the PMDETAIL inventory have burn areas that exceed the area of a 12-km grid cell.

Grid Resolution	km^2	Acres
36	1296	320,249
12	144	35,583
4	16	3,953

ENVIRON developed an approach for processing the DEASCO3 fires for input to CAMx that has been adopted for the PMDETAIL fires. A Fortran utility compares the daily acreage of each fire event in the inventory to the grid cell area. If the fire acreage is greater than 20% of the grid cell area, the fire emissions are distributed into a matrix of cells surrounding the grid cell that contains the fire centroid.

A similar approach is needed for processing these data for input to CMAQ.

Proposed Action

Modify the SMOKE program Grdmat to use a spatial interpolation approach to distribute PMDETAIL fires based on the ratio of the fire size to the model grid resolution. This modification would be applicable to all fire inventories that use the Air Sciences/WRAP fires inventory methodology. It would also be applicable for preparing these inventories for both CMAQ and CAMx, avoiding the need for a utility that is separate from the conventional emissions processing sequence.

Fire Plume Rise

The Air Sciences/WRAP fires inventory methodology prescribes plume characteristics to each fire event in the inventory based on the fire type and fire size. Three plume parameters, Smoldering Fraction (LAY1F), Plume Bottom (PBOT), and Plume Top (PTOP), describe the vertical extent and distribution of the emissions. These parameters are used to allocate the emissions into the vertical layer structure of air quality models.

Plume rise for air quality models is performed either offline or inline. The offline approach develops 3-D emissions files using an emissions processor (SMOKE). SMOKE combines the inventory plume parameters with a meteorology data file that defines the air quality modeling vertical layer structure to distribute the emissions into the model layers. The air quality model then reads the vertically distributed emissions directly into the layers in which they are emitted. The drawback to this approach is that the 3-D emissions files are a data storage and transfer burden because they are large files.

In the inline approach, the emissions processor passes the plume rise parameters from the inventory to the air quality model. The matching of the vertical plume extent to the model layers is done "inline" at each time-step that the model reads the emissions files. There is a performance penalty for this additional computation, but it is an advantageous trade-off to the file size requirements of the 3-D emissions files.

CAMx only has an inline plume rise option for emissions, both for fire and non-fire sources. The emissions processor mentioned above that ENVIRON developed for calculating the horizontal distribution of the fires also outputs the plume parameters in a CAMx-ready format. For the WestJumpAQMS project ENVIRON instrumented CAMx to read the Air Science/WRAP fire plume parameters from these emissions files and use these parameters for the inline vertical distribution of the fire emissions.

CMAQ has options for both offline and inline plume rise calculations. The CMAQ approach for inline plume rise for fires is directly adopted from SMOKE and requires that the inventories include the fire size (acres), fuel loads (tons/acre) and optionally heat release (BTU/hr). SMOKE is used to prepare the inline emissions input files for these types of fire inventories. SMOKE can also process the Air Sciences/WRAP fire inventories, although currently the only option for this inventory format is to create 3-D emissions files.

With the existing versions of SMOKE and CMAQ there are two options for using the Air Sciences/WRAP fires inventories in CMAQ.

1. Process the inventories as 3-D sources through SMOKE and combine these with the rest of the low-level and inline point sources (i.e. EGUs, non-EGUs, O&G point) in CMAQ
2. Obtain the fire size and fuel loading data for the PMDETAIL inventory from the Air Sciences database and process the emissions in SMOKE and CMAQ using the existing plume rise methodology

The advantage of option 1 is that it preserves the fire plume structures of the Air Sciences/WRAP fire inventories and does not require any modification to the existing tools; the disadvantage is that it requires using 3-D emissions files. The sizes of the 3-D emissions files are a particular concern for the current and future 3SAQS/Western modeling platforms given the size of the 12-km and 4-km nested modeling domains. The advantage of option 2 is that it takes advantage of available data and tools and will not require any modifications to CMAQ. Option 2 also requires less effort for data processing and management than option 1. The disadvantage of option 2 is that it overrides the Air Sciences/WRAP fire inventory plume information with a more crude plume rise approximation that maps the fire heat flux to plume buoyancy.

Proposed Action

Modify the SMOKE program Elevpoint to pass the Air Sciences/WRAP fire plume characteristics (LAY1F, PTOP, PBOT) through to the inline and stack groups emissions files required for computing inline plume rise with CMAQ. Modify CMAQ to read these parameters and distribute the fire emissions to the vertical model layers inline.

UNC-IE estimates that we can complete the SMOKE and CMAQ modifications described in the proposed actions in the next month. The effort will require time for development and testing from a senior software engineer and cost about $8,000.